Heatsinks for DIY LED lamps

coolbreez1

coolbreez1

Well-Known Member
The drivers are Meanwell LPF-90-42, I should have mounted them with a bit more space between them, but they don't don't seem to be going over 145 degrees.

The LED COBs seem to stay around 100 degrees, or 15-20 above ambient.
 
SupraSPL

SupraSPL

Well-Known Member
Here is an interesting test comparing the temp droop of these 4:
CXA3070 3K AB on stock Alpine 11 @ 2.23A
CXA3070 3K AB on stock Rosewill RCX-Z1 @ 2.23A
Vero 29 3K on stock Rosewill RCX-Z1 @ 2.23A
CXA3070 3K Z4 on Intel D60188-001 @ 2.23A
DSC07744a.jpg
--------------------------------------------------------------------
CXA3070 3K AB on stock Alpine 11 @ 2.23A
no fan - 53C - 10.4% temp droop (stopped test, temp rising fast)
5V - 40C - 5.43% temp droop
7.5V - 36C - 3.9% temp droop
9V - 35C - 3.56% temp droop
------------------------------------------------------------------------
CXA3070 3K AB on stock Rosewill RCX-Z1 @ 2.23A
no fan - 53C - 9.72% temp droop (stopped test, temp rising fast)
5V - 39C - 4.63% temp droop
7.5V - 36C - 4% temp droop
9V - 35C - 3.72% temp droop
-----------------------------------------------------------------------
Vero 29 3K on stock Rosewill RCX-Z1 @ 2.23A
no fan - 51C - 7% temp droop (I stopped the test, temp rising fast)
5V - 37C - 3.57% temp droop
7.5V - 34C - 3.2% temp droop
9V - 33C - 3.02% temp droop
-----------------------------------------------------------------------
CXA3070 3K Z4 on round Intel w copper base @ 2.23A
no fan - 56C - 9.34% temp droop (stopped test, temp rising fast)
4.11V - 38C - 3.61% temp droop
5V - 35C - 2.29% temp droop
7.5V - 37C - 3.51% temp droop
9V - 35C - 2.91% temp droop

So the results with the fan power and temp droop factored in (tomorrow)
CXA/Alpine 5V -
CXA/Alpine 9V -
CXA/Rosewill 5V -
CXA/Rosewill 7.5V -
CXA/Rosewill 9V -
Vero Rosewill 5V -
Vero Rosewill 7.5V -
Vero/Rosewill 9V -
CXA/Intel 4.1V -
CXA/Intel 5V -
CXA/Intel 7.5V -
CXA/Intel 9V -

In summary the Vero outperformed the CXA by almost 1%, as we would expect considering its much higher current limit and larger surface area, very nice!. The Intel has strange fan characteristics but it performed amazingly well at 5V and somehow outperformed every other setup. The blow through design and large copper pad must have really made the difference.

This test is not quantitative and they cannot be compared because of slightly differing angles and the Veros redder spectra, but it may be worth mentioning the actual pulsed lux measurments of each anyway because they were all taken from the same distance.
CXA AB Alipne 11 - 1124
CXA AB Rosewill - 1101
Vero29 Rosewill - 1093
CXA Z4 Intel - 996

Also worth mentioning, the Vero 29 Vf was much higher than the "typical" figures in the PDF chart and it dropped almost twice from cold to warmed up versus the CXA3070. The Vero29 Vf was only slightly lower than the CXA3070s at 2.23A.
 
Last edited:
AquariusPanta

AquariusPanta

Well-Known Member
SupraSPL said:
Here is an interesting test comparing the temp droop of these 4:
CXA3070 3K AB on stock Alpine 11 @ 2.23A
CXA3070 3K AB on stock Rosewill RCX-Z1 @ 2.23A
Vero 29 3K on stock Rosewill RCX-Z1 @ 2.23A
CXA3070 3K Z4 on Intel D60188-001 @ 2.23A
View attachment 3310183
--------------------------------------------------------------------
CXA3070 3K AB on stock Alpine 11 @ 2.23A
no fan - 53C - 10.4% temp droop (stopped test, temp rising fast)
5V - 40C - 5.43%
7.5V - 36C - 3.9%
9V - 35C - 3.56%
------------------------------------------------------------------------
CXA3070 3K AB on stock Rosewill RCX-Z1 @ 2.23A
no fan - 53C - 9.72% temp droop (stopped test, temp rising fast)
5V - 39C - 4.63%
7.5V - 36C - 4%
9V - 35C - 3.72%
-----------------------------------------------------------------------
Vero 29 3K on stock Rosewill RCX-Z1 @ 2.23A
no fan - 51C - 7% temp droop (I stopped the test, temp rising fast)
5V - 37C - 3.57% temp droop
7.5V - 34C - 3.2% temp droop
9V - 33C - 3.02% temp droop
-----------------------------------------------------------------------
CXA3070 3K Z4 on round Intel w copper base @ 2.23A
no fan - 56C - 9.34% temp droop (stopped test, temp rising fast)
4.11V - 38C - 3.61%
5V - 35C - 2.29%
7.5V - 37C - 3.51%
9V - 35C - 2.91%

So the results with the fan power and temp droop factored in (tomorrow)
CXA/Alpine 5V -
CXA/Alpine 9V -
CXA/Rosewill 5V -
CXA/Rosewill 7.5V -
CXA/Rosewill 9V -
Vero Rosewill 5V -
Vero Rosewill 7.5V -
Vero/Rosewill 9V -
CXA/Intel 4.1V -
CXA/Intel 5V -
CXA/Intel 7.5V -
CXA/Intel 9V -

In summary the Vero outperformed the CXA by almost 1%, as we would expect considering its much higher current limit and larger surface area, very nice!. The Intel has strange fan characteristics but it performed amazingly well at 5V and somehow outperformed every other setup. The blow through design and large copper pad must have really made the difference.

This test is not quantitative and they cannot be compared because of slightly differing angles and the Veros redder spectra, but it may be worth mentioning the actual pulsed lux measurments of each anyway because they were all taken from the same distance.
CXA AB Alipne 11 - 1124
CXA AB Rosewill - 1101
Vero29 Rosewill - 1093
CXA Z4 Intel - 996
Click to expand...
For the Intel w/copper base, you have 5V being the lowest percentage or the most efficient amongst the fan's potential range. Was that a typo/error/misreading or a phenomena? :-P

And yes, the intel fans that I have are strange birds. Maybe yours are stranger though.
 
bicit

bicit

Well-Known Member
SupraSPL said:
Also worth mentioning, the Vero 29 Vf was much higher than the "typical" figures in the PDF chart and it dropped almost twice from cold to warmed up versus the CXA3070. The Vero29 Vf was only slightly lower than the CXA3070s at 2.23A.
Click to expand...
What was the Vf on the vero 29 in operation?
 
coolbreez1

coolbreez1

Well-Known Member
SupraSPL said:
How high above the canopy do you run them? Probably very good reflective walls but maybe benefit from some DIY reflectors? I figure it at 485W dissipation, 38% efficient (typical) or 185 PAR W, that is some serious output for a bath tub bongsmilie
Click to expand...
They are Vero 29, 4000k, I thought this would be a good compromise for starts/veg. I was a bit behind on the light so it went over some very leggy starts, as such it is way above the plants at the time being, like 6 feet.IMG_1187.JPG
As you can see in the picture, they were pretty sad little starts....And And here is another picture of them at about two weeks under the lights.
IMG_1204.JPG
Pretty much every start, even a few I had very little hope for were saved by the light, and the ones a bit closer to the light are growing noticeably faster, note the two in the left and right corners, they have been there the entire time.

I have been considering some reflectors... mostly just so I don't blind my self walking into the room :) I am thinking that now that the ladies seem well adjusted raising them up closer to the light, I could lower the light because it is on chains for that very reason, but then I would have to worry about hitting it with my head.

I am starting to consider my next set of lights for flower, likely the Vero29 2700k, but I need to look into those Citizen LEDs now as well, with the much higher red peak going upwards of like 760nm.
 
Last edited:
Positivity

Positivity

Well-Known Member
Good stuff supra! i didnt think the arctic could handle that high a load. That will make it easy for peeps who want to run a little harder.

Starting to notice a mass vs airflow relationship..

I wonder how the measuring techniques differ though. A heat gun versus a thermocouple? Would they give the same measurements? Probably really close..
 
Dloomis514

Dloomis514

Well-Known Member
SupraSPL said:
Here is an interesting test comparing the temp droop of these 4:
CXA3070 3K AB on stock Alpine 11 @ 2.23A
CXA3070 3K AB on stock Rosewill RCX-Z1 @ 2.23A
Vero 29 3K on stock Rosewill RCX-Z1 @ 2.23A
CXA3070 3K Z4 on Intel D60188-001 @ 2.23A
View attachment 3310183
--------------------------------------------------------------------
CXA3070 3K AB on stock Alpine 11 @ 2.23A
no fan - 53C - 10.4% temp droop (stopped test, temp rising fast)
5V - 40C - 5.43% temp droop
7.5V - 36C - 3.9% temp droop
9V - 35C - 3.56% temp droop
------------------------------------------------------------------------
CXA3070 3K AB on stock Rosewill RCX-Z1 @ 2.23A
no fan - 53C - 9.72% temp droop (stopped test, temp rising fast)
5V - 39C - 4.63% temp droop
7.5V - 36C - 4% temp droop
9V - 35C - 3.72% temp droop
-----------------------------------------------------------------------
Vero 29 3K on stock Rosewill RCX-Z1 @ 2.23A
no fan - 51C - 7% temp droop (I stopped the test, temp rising fast)
5V - 37C - 3.57% temp droop
7.5V - 34C - 3.2% temp droop
9V - 33C - 3.02% temp droop
-----------------------------------------------------------------------
CXA3070 3K Z4 on round Intel w copper base @ 2.23A
no fan - 56C - 9.34% temp droop (stopped test, temp rising fast)
4.11V - 38C - 3.61% temp droop
5V - 35C - 2.29% temp droop
7.5V - 37C - 3.51% temp droop
9V - 35C - 2.91% temp droop

So the results with the fan power and temp droop factored in (tomorrow)
CXA/Alpine 5V -
CXA/Alpine 9V -
CXA/Rosewill 5V -
CXA/Rosewill 7.5V -
CXA/Rosewill 9V -
Vero Rosewill 5V -
Vero Rosewill 7.5V -
Vero/Rosewill 9V -
CXA/Intel 4.1V -
CXA/Intel 5V -
CXA/Intel 7.5V -
CXA/Intel 9V -

In summary the Vero outperformed the CXA by almost 1%, as we would expect considering its much higher current limit and larger surface area, very nice!. The Intel has strange fan characteristics but it performed amazingly well at 5V and somehow outperformed every other setup. The blow through design and large copper pad must have really made the difference.

This test is not quantitative and they cannot be compared because of slightly differing angles and the Veros redder spectra, but it may be worth mentioning the actual pulsed lux measurments of each anyway because they were all taken from the same distance.
CXA AB Alipne 11 - 1124
CXA AB Rosewill - 1101
Vero29 Rosewill - 1093
CXA Z4 Intel - 996

Also worth mentioning, the Vero 29 Vf was much higher than the "typical" figures in the PDF chart and it dropped almost twice from cold to warmed up versus the CXA3070. The Vero29 Vf was only slightly lower than the CXA3070s at 2.23A.
Click to expand...
Would I be making a wrong conclusion to say any heat sink tested at 5v cools well enough, or its that an over simplification? Seems like the run of the mill cpu cooler does a dam fine job at five volts.

Thanks for the sharing your work!
 
Dloomis514

Dloomis514

Well-Known Member
I have been looking for old Cpu coolers thinking they could be had cheap. But looks like they don't get salvaged. Seems like an opportunity there
 
SupraSPL

SupraSPL

Well-Known Member
Yes I think that is a good generalization. The only time we have seen less than optimal performance from 5V was the Alpine 11 with CXA @ 2.23A. But the Alpine fan is low RPM so at 5V uses only .555W. So that exposes the limitations where 7.5V becomes a better option.
 
SupraSPL

SupraSPL

Well-Known Member
@stardustsailor from what I am seeing, I think it would be relatively easy and worthwhile to get the green/yellow zone in those pics down to 35C. When I take the heatsink temps I take them from the hottest part I can get repeatable numbers from (where the photon emission will not affect the reading).
 
Last edited:
SupraSPL

SupraSPL

Well-Known Member
Here are the results of the last test. This time I used actual efficiencies, as best as I could estimate and I took into account the decrease in dissipation due to temp lowering the Vf. What it shows is that with a relatively large heat load (COBS at 2.23A), 5V is still the best option for fan power, except in the case of the Alpine 11 which has a low RPM fan and cooled slightly more efficiency with 7.5V.

It shows that the Arctic 11 and Rosewill seem to be evenly matched (CXA vs CXA)

And finally, it shows the Vero29 3K provides 5.3% more light than the CXA3070 3K AB at 2.23A, partly because the Vero handled the heat better and partly because of current droop at high currents (we are trusting in "typical" figures and LER of 315 for the 3K Vero that SDS provided). Using 321 LER the Vero creates 3.38% more light than the CXA.
Heatsinks at 2.23A.png
 
Last edited:
bicit

bicit

Well-Known Member
SupraSPL said:
And finally, it shows the Vero29 3K provides 5.3% more light than the CXA3070 3K AB at 2.23A, partly because the Vero handled the heat better and partly because of current droop at high currents (we are trusting in "typical" figures and LER of 315 for the 3K Vero that SDS provided).
View attachment 3310660
Click to expand...
Thanks for doing this testing supra, really important stuff to know. I'm sure you're already planning on it, but do you anticipate seeing how the Vero and CXA stack up at lower currents?
 
stardustsailor

stardustsailor

Well-Known Member
SupraSPL said:
Here are the results of the last test. This time I used actual efficiencies, as best as I could estimate and I took into account the decrease in dissipation due to temp lowering the Vf. What it shows is that with a relatively large heat load (COBS at 2.23A), 5V is still the best option for fan power, except in the case of the Alpine 11 which has a low RPM fan and cooled slightly more efficiency with 7.5V.

It shows that the Arctic 11 and Rosewill seem to be evenly matched (CXA vs CXA)

And finally, it shows the Vero29 3K provides 5.3% more light than the CXA3070 3K AB at 2.23A, partly because the Vero handled the heat better and partly because of current droop at high currents (we are trusting in "typical" figures and LER of 315 for the 3K Vero that SDS provided).
View attachment 3310660
Click to expand...
Well ...
The spectral graph of the 'old' datasheet of Vero29 for the 3K-Ra80 ,reveals a LER of 315 ,indeed.
old.JPG

But at the new/updated datasheet of Vero29 ,the spectral graph for the same 3K-Ra80 gives a LER of
321 at the particular COB .
new.JPG

Not my mistake .
Seems that Vero has ~10% of quantum flux at the 400-499nm ( aka "blue" ) range.
Same 602-603 nm red "peak" ,as with the 3K-80 CXA ...
And still a tad more reds ...(in fact ,slightly wider phoshor emission at general,over the CXA3070 3K-Ra80 )

The main reason ,behind the Veros ,slightly better efficiency figures,is this :
Simply the Vero 29 is a more powerful COB ,than the CXA3070.

Hardly assuming that both technologies incorporated ( Ga on Si & SiC ) have as a result ,
the manufactured blue dies to be performing about the same (>50% radiometric efficiency ).

CXA 3070 has about 132 blue diodes underneath it's LES .(if I remember correctly ... )
While the Vero29 has over 150 blue diodes ,underneath the LES...
vero chips.jpg
Which means ....
For the same driving current ....
Vero29 utilises more diodes than the CXA 3070...
Thus lower driven ,than the diodes of the CXA3070 ..
......
Thus ,having better radiometric efficiencies,of course ...

Further more..
More blue light escaping the LES...
Means-most probably - that less phosphor load is used (quantity ) ...
Thus less losses..

Just saying .

Cheers.
:peace:
 
Last edited:
You must log in or register to reply here.
Top